Le 11.11.2014 22:53, Miles Fidelman a écrit :
On a broader note, Debian, Linux, *nix in general, and FOSS software are a complex and highly-interdependent ecosystem. Yes some people just take, but an awful lot of us contribute in various ways, in various places, to the overall ecosystem - be it writing upstream code, libraries, documentation, providing training, doing policy work(can you say EFF), crafting open-source licenses, providing support invarious forms. The gnu tools, glibc, the kernel - without those,there would be no Debian or other distributions. Arguably, without theGPL, there wouldn't be a lot of FOSS software. EFF goes out and fights legal battles to protect the ecosystem. An awful lot of code depends on Apache, MySQL, SQL Lite, and so forth. And it goes on. The pieces are highly interdependent, and in many cases, a contribution to one project, or activity, benefits many others.
While I generally agree in your ideas, I disagree that all, or most, pieces are that interdependent (but, some are, yes. I usually try to avoid those, thought, because it's a bad idea to put all yours eggs in the same basket. I favor portable tools, and when I contribute to something, my contributions never goes in non-portability direction. Never, except when I do not know it :) ).
My reason is that, excepted when you start your software based on non-standard, non-portable tools, you can replace parts of the ecosystem with other tools. You speak about mysql/sqlite for example. Stuff which relies on specificities of mysql (for example) are not easy to port, indeed, but lot of things which just rely on SQL can be ported quite easily from a SQL engine to another. It's one of the reason for which I generally tend to limit myself to pure SQL when I need stuff of this kind. For sqlite, it's harder, because it happens that it's the easiest and probably more efficient SQL engine able to avoid bothering the user with painful rights problems and complex setup. Now, firebird is also able to run in an embedded mode.
In facts, imho the FOSS' power comes from that fact: when there is enough need for a technology, alternatives spawn, because there is always someone to disagree about how things are done. And when there are alternatives, choice, pieces of the choice leads other pieces to improvement (a good example here is clang vs gcc). The only thing which comes into my mind which does not respect that point is... Xorg, for which I does not know about any alternative (which does not mean that there is no alternative). Probably because it's very, very complex, probably too much, and I've heard that this complexity is the reason behind wayland. And with wayland, we can hope that in few years there could be other implementations of the protocol. At first, I wanted to speak about GCC as another unique software without alternatives, and the glibc (which is a part of GCC). However, I am not in FOSS world since enough time to know how it was before, and nowadays, there is clang, which is also far better for my uses.
In short: bazaar includes various versions of the same kind of stuff.The fact that pieces of a bazaar (which can be cathedrals, why not?) can be replaced by other is a strength. But, in the end, yes, contributing to some pieces can imply improvements of the whole picture. And, obviously, code is not the only way to contribute to a project. On small projects with only few devs, even simply showing them interest help.
And before someone says it, yes, for systemD there are inter-operable alternatives. At least one: uselessd (remember: I said that there is always someone to disagree ;) ).